@Erik_Cederberg , let’s raise the the bar a little 
I quickly made a dual extrusion version of +Maker’s Muse tolerance test. Printed on the Ultimaker Mark2 magnetic tool changer, 0.1 layer, 2 x 0.4 nozzle. Working down to 0.1 mm (with some friction and it took some minutes of wiggling to get this one loose). https://youtu.be/xhzBjBGG1JY
That detail is on point!
I’m still surprised from time to time how well this simple printed kinematic coupling (just € 2 for the 6 magnets) performs.
Haha, nice one @Markus_Seidt , you know I love the Mark2 
But since you are asking, I will try it on the UM3 overnight, can you send me the dual file so I don’t have to do it myself? 
I feel like there’s some sort of X/Y compensation going on here, the gaps look way larger than on the file or anyone else’s prints.
@Thomas_Sanladerer , nope, standard Cura 2.5 settings. There is an outer wall inset setting, which for the Mark2 defaults to 0.025 mm. I think 2.5/100 mm is in the same ball park as over-/underextrusion effects. The 0.05 test is fused, 0.1 has quite some friction, 0.15 and 0.2 show virtually no play, noticeable vertical play starts at 0.3 mm. You could just build a Mark2 magnetic tool changer and try on our own
That print looks amazing
@Markus_Seidt I think @Thomas_Sanladerer has a point. The gaps definitely look larger than any other print off this test I’ve seen. The outer wall inset you mentioned might have a larger impact than I would think from the value you stated. But the gaps really looking bigger than they should.
@Sven_Muller , colors and lighting (shadows) make a huge difference in visual appearance and can be misleading. Look at the tiny white lines, where I try to measure and compare as good as possible. To my experience the key for printing this part well is temperature and speed.
missing/deleted image from Google+
As Angus already mentioned, most likely the parts will fuse close to the middle where the overhangs have the smallest diameter. An angle of like 40° is nothing serious but with 8 mm in diameter and 50 mm/s speed the nozzle will heat up every spot there already twice a second. Cooling air can only reach the top because the outer surface is covered by the other part. This is where you start getting surface imperfections due to local overheating, even with 40° overhangs. I used just 2 perimeter lines to avoid a third pass with additional reheating, printed with reasonably low temp (200°C extrudr GreenTec) and with just 30-40 mm/s for the walls.
I call this black and red print BS, playing with X Y compensations!!! The gaps are way too large.
I do trust @Markus_Seidt if he says there is nothing funny done with the settings. I did not have time to start a print last night on the UM3, but will get back here as soon as I have had time to print it
For the people interested in serious analysis. 15.98 (vs 16.0), and 18.89 (vs - hard to tell exactly from the stl - 18.9)
missing/deleted image from Google+
missing/deleted image from Google+
@Markus_Seidt That gray one in the pic is the one I printed. All I changed was using a 0.2 tip. Took 17 hours but came out beautiful. ^^
@Kura_kuea Wow, I think 0.08 x 0.2 mm with dual extrusion is beyond the time I’d be willing to wait (+24 hrs). Currently the dual print takes 6.5 hrs. However I’m not sure that a 0.2 mm nozzle makes that much difference. The challenge is printing tiny gaps, not tiny structures nor super sharp corners. 0.2 gives more precision but the overhangs might become more difficult. Yesterday I had my first ever print with PETG fused at 0.1 mm . As expected right in the middle where diameters and gaps are minimal.
@Markus_Seidt I love PETG, but it is more “sticky” then PLA. This test is more about how well tuned you have movements and extrusion, both of which be come much harder with smaller tips and layers. Mine was good down to the 0.15, 0.1 turned but broke out when I forced it. ^^
missing/deleted image from Google+
Next attempt with extrudr PETG. Still not satisfied with extrusion consistency, PETG doesn’t like speed changes very much (at least with the Ultimaker extrusion train)
missing/deleted image from Google+
@Markus_Seidt Not at all. I found slow and steedy works nicely around 20mm/s for me.